U.S. patent number 7,434,114 [Application Number 11/326,622] was granted by the patent office on 2008-10-07 for method of compensating for a byte skew of pci express and pci express physical layer receiver for the same.
This patent grant is currently assigned to Samsung Electronics, Co., Ltd.. Invention is credited to Soon-Bok Jang, Young-Gyu Kang.
United States Patent |
7,434,114 |
Jang , et al. |
October 7, 2008 |
Method of compensating for a byte skew of PCI express and PCI
express physical layer receiver for the same
Abstract
A method of compensating for a byte skew of a PCI Express bus,
the method including determining whether received data are in a
training sequence or not, setting an alignment point corresponding
to each of the lanes based on a comma symbol included in the
training sequence when the received data are in the training
sequence, and shifting the alignment point by reflecting an
addition or a removal of a skip symbol on the received data through
each of the four lanes when the received data are not in the
training sequence. Therefore, the byte skew of the PCI Express bus
may be effectively compensated for despite the addition or the
removal of the skip symbol.
Inventors: |
Jang; Soon-Bok (Suwon-si,
KR), Kang; Young-Gyu (Hwaseong-si, KR) |
Assignee: |
Samsung Electronics, Co., Ltd.
(Gyeonggi-do, KR)
|
Family
ID: |
36654692 |
Appl.
No.: |
11/326,622 |
Filed: |
January 7, 2006 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20060156083 A1 |
Jul 13, 2006 |
|
Foreign Application Priority Data
|
|
|
|
|
Jan 10, 2005 [KR] |
|
|
10-2005-0001995 |
|
Current U.S.
Class: |
714/700; 714/43;
714/709; 714/712; 714/707; 714/701; 714/56; 714/715; 714/724;
714/735; 714/744; 714/799; 714/25 |
Current CPC
Class: |
G06F
13/423 (20130101) |
Current International
Class: |
G06K
5/04 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
11074945 |
|
Mar 1999 |
|
JP |
|
11341102 |
|
Dec 1999 |
|
JP |
|
20020044061 |
|
Feb 2002 |
|
JP |
|
20030090954 |
|
Dec 2003 |
|
KR |
|
Other References
PCI-SIG, techsupp@pcisig.com, "PCI Express Base Specification
Revision 1.0", Apr. 29, 2002, pp. 158-160. cited by
examiner.
|
Primary Examiner: Trimmings; John P
Attorney, Agent or Firm: F. Chau & Associates, LLC
Claims
What is claimed is:
1. A method of compensating for a byte skew of a PCI Express bus,
the method comprising: determining whether or not a plurality of
data streams are in a training sequence, each of the plurality of
data streams being received through a respective lane; setting an
alignment point for each respective lane based on a comma symbol
included in the corresponding data stream received through the
respective lane, when the received data streams are in the training
sequence; and shifting an alignment point by adding or removing a
skip symbol in the corresponding data stream, when the received
data streams are not in the training sequence, wherein shifting an
alignment point comprises: determining whether or not the skip
symbol is added into the corresponding received data stream for
each of the lanes, or whether or not the skip symbol is removed
from the corresponding received data stream for each of the lanes;
shifting each of the alignment points so that the corresponding
received data stream is less delayed by shifting the alignment
points of the other lanes except for the corresponding lane wherein
the other received data streams except for the corresponding
received data stream are more delayed when the corresponding
received data stream cannot be less delayed, when the skip symbol
is added into the corresponding received data stream; and shifting
each of the alignment points so that the corresponding received
data stream is more delayed by shifting the alignment points of the
other lanes except for the corresponding lane wherein the other
received data streams except for the corresponding received data
stream are less delayed when the corresponding received data stream
cannot be more delayed, when the skip symbol is removed from the
corresponding received data stream.
2. The method of claim 1, wherein a number of the lanes is
four.
3. The method of claim 1, wherein setting an alignment point
comprises: determining whether or not there exists a lane where the
comma symbol is detected; determining whether or not comma symbols
are detected in all of the lanes during a predetermined time
period, when there exists a lane where the comma symbol is
detected; and setting each of the alignment points based on each of
timings when the corresponding comma symbol is detected, when the
comma symbols are detected in all of the lanes during the
predetermined time period.
4. The method of claim 3, wherein determining whether or not the
comma symbols are detected in all of the lanes during the
predetermined time period, comprises: starting a timer when there
exists the lane where the comma symbol is detected; determining
whether or not the comma symbols are detected in all of the lanes;
and determining whether or not the timer is expired.
5. A PCI Express physical layer receiver comprising: a deserializer
configured to convert each of a plurality of serial data streams
received through respective lanes to respective 10-bit parallel
data streams; a PHY Interface for the PCI Express Architecture
(PIPE) configured to perform a bit alignment process on each of the
10-bit parallel data streams based on comma symbols, configured to
convert each of the 10-bit parallel data streams to respective
8-bit parallel data streams, and configured to add a skip symbol to
the 8-bit parallel data stream or remove a skip symbol from the
8-bit parallel data stream based on a status of the corresponding
received data stream; and a skew compensator configured to set
alignment points corresponding to each of the lanes based on a
corresponding comma symbol included in the corresponding 8-bit
parallel data stream, and configured to compensate for a byte skew
between the 8-bit parallel data streams by shifting each of the
alignment points based on the addition or removal of the skip
symbol in the corresponding 8-bit parallel data stream, wherein the
skew compensator is configured to shift each of the alignment
points so that the corresponding 8-bit parallel data stream is less
delayed when the skip symbol is added into the corresponding 8-bit
parallel data stream, and configured to shift the alignment points
so that the corresponding 8-bit parallel data stream is more
delayed when the skip symbol is removed from the corresponding
8-bit parallel data stream.
6. The PCI Express physical layer receiver of claim 5, wherein a
number of the lanes is four.
7. The PCI Express physical layer receiver of claim 5, wherein the
skew compensator receives information on the addition or the
removal of the skip symbol from the PIPE.
8. The PCI Express physical layer receiver of claim 5, wherein the
skew compensator is configured to determine whether or not there
exists the lane where the corresponding comma symbol is detected,
configured to determine whether or not the comma symbols are
detected in all of the lanes during a predetermined time period,
when there exists the lane where the comma symbol is detected, and
configured to set each of the alignment points based on each of
timings when the corresponding comma symbols is detected, when the
comma symbols are detected in all of the lanes during the
predetermined time period.
9. The PCI Express physical layer receiver of claim 8, wherein the
skew compensator comprises a timer for determining whether or not
the comma symbols are detected in all of the lanes during the
predetermined time period.
10. The PCI Express physical layer receiver of claim 9, wherein the
timer is configured to be started when there exists the lane where
the corresponding comma symbol is detected, and configured to be
reset when the comma symbols are not detected in all of the lanes
during the predetermined time period.
11. The PCI Express physical layer receiver of claim 5, wherein the
skew compensator is configured to shift the alignment points of the
other lanes except for the corresponding lane wherein the other
8-bit parallel data stream except for the corresponding 8-bit
parallel data stream are more delayed when the corresponding 8-bit
parallel data stream cannot be less delayed.
12. The PCI Express physical layer receiver of claim 5, wherein the
skew compensator is configured to shift the alignment points of the
other lanes except for the corresponding lane wherein the other
8-bit parallel data streams except for the corresponding data
stream are less delayed when the corresponding 8-bit parallel data
stream cannot be more delayed.
13. A method of compensating for a byte skew of a PCI Express bus,
the method comprising: determining whether or not a plurality of
data streams are in a training sequence, each of the plurality of
data streams being received through a respective lane; when the
received data streams are in the training sequence, setting an
alignment point for each lane based on a comma symbol included in
the corresponding data stream received through the corresponding
lane, wherein setting an alignment point comprises determining
whether or not there exists a lane where the comma symbol is
detected; setting each of the alignment points based on each of
timings when the corresponding comma symbol is detected; and when
the received data streams are not in the training sequence,
shifting an alignment point by adding or removing a skip symbol in
the corresponding data stream, wherein shifting each of the
alignment points comprises: determining whether or not the skip
symbol is added into the corresponding received data stream for
each of the lanes, or whether or not the skip symbol is removed
from the corresponding received data stream for each of the lanes;
and when the skip symbol is added into the corresponding received
data stream, shifting each of the alignment points of the other
lanes except for the corresponding lane wherein the other received
data streams except for the corresponding received data stream are
more delayed when the corresponding received data stream cannot be
less delayed.
14. The method of claim 13, wherein when the skip symbol is removed
from the corresponding received data stream, shifting each of the
alignment points of the other lanes except for the corresponding
lane wherein the other received data streams except for the
corresponding received data stream are less delayed when the
corresponding received data stream cannot be more delayed.
15. The method of claim 13, further comprising, when there exists a
lane where the comma symbol is detected, determining whether or not
comma symbols are detected in all of the lanes during a
predetermined time period, wherein determining whether or not the
comma symbols are detected in all of the lanes during the
predetermined time period comprises: starting a timer when there
exists the lane where the comma symbol is detected; determining
whether or not the comma symbols are detected in all of the lanes;
determining whether or not the timer is expired; and setting each
of the alignment points based on each of timings when the comma
symbols are detected in all of the lanes during the predetermined
time period.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
This application claims priority to Korean Patent Application No.
2005-1995 filed on Jan. 10, 2005 in the Korean Intellectual
Property Office, the entire contents of which are incorporated
herein by reference.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to PCI (Peripheral Component
Interconnect) Express, and more particularly to a method of
compensating for a byte skew of PCI Express and a PCI Express
physical layer receiver for the same.
2. Description of the Related Art
PCI Express was introduced in 2002 to overcome speed limitations of
conventional PCI, so as to match the speed of current CPUs. The
conventional PCI signals were easily distorted due to a parallel
transmission interference. Accordingly, it is difficult to increase
the clock frequency of a conventional PCI bus. A PCI Express bus,
which employs a serial transmission manner, is capable of
increasing the clock frequency and is capable of reducing a bus
size.
PCI Express, formerly known as 3rd Generation I/O (3GIO), has
replaced conventional PCI in a wide variety of fields. PCI Express
features low-voltage differential signaling (LVDS), a packet-based
data transmission protocol and so on. PCI Express supports a dual
simplex type bus. The dual simplex bus includes a pair of one
directional data buses, one bus is used for transmitting data and
the other bus is used for receiving the data. Due to the LVDS, the
PCI Express bus uses a 4-wire interface per lane. As a result, the
PCI Express bus uses more wires per data bit than that of the
conventional data bus. Also, a message-based protocol and an
embedded clocking of the PCI Express bus may contribute to omitting
usage of various data control signals required for an interface
process of the conventional data bus. The PCI Express bus may
include a maximum of 32 lanes. Generally, when a multi-lane bus,
such as the PCI Express bus is used, a transmitter transmits
divided data to each of the lanes so as to improve transmission
efficiency. For example, when the number of the lanes is 4, the
data to be transmitted to a receiver is divided to each of 32-bit
data packets and the 32-bit data packets are divided to each of
8-bit data packets. Each of the 8-bit data packets is transmitted
through each of the four lanes at the same time. The receiver
aligns the received data through the wire by compensating for a bit
skew between the received data using a comma symbol, and then,
aligns the first aligned received data for which the bit skew is
compensated, by compensating for a byte skew between the first
aligned received data. The bit skew compensation deals with the
skew within 8 bits between the received data through each of the
lanes bit by bit. Because each of the lanes has a different
transmission delay in multi-lane systems, the skew occurs between
the lanes. When a transmission speed is relatively low, the skew
between the lanes occurs within 8 bits. Accordingly, data reception
is effectively performed by compensating for only the bit skew
between the lanes. However, in high-speed data transmission devices
such as the PCI Express bus, a difference of the transmission delay
between the lanes may occur out of the 8-bit range. Accordingly,
the high-speed data transmission devices such as the PCI Express
bus require a byte skew compensation, as well as the bit skew
compensation. The byte skew compensation deals with the byte skew
between the received data through each of the lanes byte by
byte.
FIG. 1 is a block diagram illustrating a data transmission of a
conventional PCI Express bus.
Referring to FIG. 1, the PCI Express bus transmits data using four
lanes LANE 0 through LANE 3. A transmitter of the PCI Express bus
divides 32-bit data 110 into four 8-bit data 111, 112, 113 and 114.
Then, the transmitter transmits the four-data 111, 112, 113 and 114
through each of the lanes LANE 0, LANE 1, LANE 2 and LANE 3,
respectively. Each of the four 8-bit data 111, 112, 113 and 114 is
transmitted to a wire through an 8-bit/10-bit encoder (not shown)
and a serializer (not shown) bit by bit. Other 32-bit data 120,
130, 140, 150 and 160 are respectively divided into four 8-bit
data. Then, the divided four 8-bit data are transmitted through
each of the lanes. Each of the lanes includes differential wires
for transmission and reception, and thus, may include a total of 4
wires.
As shown in FIG. 1, when data are transmitted through each of the
four lanes, each of the four lanes takes a different transmission
delay from one another. Accordingly, the data transmitted from the
transmitter are not simultaneously received by a PCI Express
receiver. Therefore, the receiver should eliminate the skew between
the received data through each of the four lanes.
FIG. 2 is a block diagram illustrating a data reception of a
conventional PCI Express bus.
Referring to FIG. 2, the PCI Express receiver decodes bit data
received through the wire using a de-serializer (not shown) to
8-bit data. As shown in FIG. 2, the bit skew and the byte skew
occur between the lanes of the received 8-bit data 211, 212, 213,
214, 221, 222, 223, 224, 231, 232, 233, 234, 241, 242, 243 and 244.
The reason for the bit skew and the byte skew occurring is that
each of the lanes takes a differential transmission delay from one
another. The PCI Express bus compensates for the bit skew between
the received 8-bit data using a comma symbol to align the received
8-bit data and generates the aligned 8-bit data 271, 272, 273, 274,
275, 276, 277 and 278. The comma symbol may include a particular
bit composition. The PCI Express bus generates 32-bit data 280
using the four 8-bit data 271, 272, 273 and 274. Each of the 8-bit
data 271, 272, 273 and 274 included in the 32-bit data 280 does not
include the bit skew, but the 32-bit data 280 includes the byte
skew. Consequently, the PCI Express bus requires a de-skew block
290 capable of compensating for the byte skew between the received
32-bit data and capable of generating 32-bit data 260 having no
byte skew.
The conventional de-skew block 290 shown in FIG. 2 compensates for
the byte skew using the comma symbol. A conventional method of
compensating for the byte skew includes a step of detecting the
comma symbol included in the received data, a step of waiting until
the comma symbols are detected in all of the lanes, and a step of
aligning the received data by delaying the received data through
other lanes based on a timing as to when the last comma symbol is
detected. However, the conventional method of compensating for the
byte skew using only the comma symbol may not appropriately
compensate for the byte skew between the received data through each
of the lanes, since skip symbols SKP are periodically transmitted
after the comma symbol thereby varying each of the alignment points
of the received data.
The PCI Express receiver includes an elastic buffer. The elastic
buffer may remove the skip symbols included in the received data to
prevent an overflow of the buffer when quantity of the received
data through a certain lane is large. In addition, the elastic
buffer may add the skip symbols to the received data to prevent an
underflow of the buffer when quantity of the received data through
a certain lane is small. When the addition or the removal of the
skip symbol occurs, it is required that the byte skew compensation
be performed by reflecting the addition or the removal of the skip
symbol in an alignment process of the received data since the skip
symbol is added or is removed to/from the received data after the
comma symbol is received.
FIG. 3 is a timing diagram illustrating a conventional method of
compensating for a byte skew using only comma symbol. Referring to
FIG. 3, a comma symbol COM and three consecutive skip symbols SKPs
are transmitted through respective lanes. According to a
conventional method of compensating for a byte skew, the comma
symbols COMs are synchronously arranged so that the data after the
comma symbol COM may be received at the same time. There is no
problem in the above method when the data of information to be
transferred are received directly after the comma symbol. The skip
symbol SKP is, however, received after the comma symbol COM, and
the skip symbol SKP is added or removed through an elastic buffer.
Therefore, the byte skew may not be compensated for, despite the
arrangement based on the comma symbols COMs.
As a result, it is desired to have a method and device which can
compensate for the byte skew regardless of an addition or a removal
of the skip symbol.
SUMMARY OF THE INVENTION
Embodiments of the present invention provide methods of
compensating for a byte skew of a PCI Express bus. These methods
may appropriately compensate for the byte skew of the PCI Express
bus by reflecting an addition or a removal of a skip symbol on a
shift operation of an alignment point.
Embodiments of the present invention also provide a PCI Express
physical layer receiver including a skew compensator. The skew
compensator may appropriately compensate for the byte skew of the
PCI Express bus by reflecting an addition or a removal of a skip
symbol on a shift operation of an alignment point.
In some embodiments of the present invention, a method of
compensating for a byte skew of PCI Express bus includes:
determining whether or not received data are in a training
sequence; setting an alignment point corresponding to each of the
lanes based on a comma symbol included in the training sequence
when the received data are in the training sequence; and shifting
the alignment point by reflecting an addition or a removal of a
skip symbol on the received data through each of the four lanes
when the received data are not in the training sequence.
In further embodiments, the setting may include determining whether
or not the lane where the comma symbol is detected exists;
determining whether or not the comma symbols are detected in all of
the lanes during a predetermined time period, when the lane where
the comma symbol is detected exists; and setting each of the
alignment points based on each of the timings when each of the
comma symbols is detected, when the comma symbols are detected in
all of the lanes during the predetermined time period.
In additional embodiments, the shifting may include determining
whether or not the skip symbol is added to the received data
corresponding to each of the lanes, or is removed from the received
data corresponding to each of the lanes; shifting the alignment
point of the corresponding received data so that the received data
are less delayed when the skip symbol is added to the received data
corresponding to each of the lanes; and shifting the alignment
point of the corresponding received data so that the received data
are more delayed when the skip symbol is removed from the received
data corresponding to each of the lanes.
When the received data of the corresponding lane cannot be less
delayed, each of the alignment points of the other lanes except for
a corresponding lane may be shifted so that the received data
corresponding to the other lanes are more delayed. In addition,
when the received data of the corresponding lane cannot be more
delayed, each of the alignment points of the other lanes except for
a corresponding lane may be shifted so that the received data
corresponding to the other lanes are less delayed.
In other embodiments of the present invention, a PCI Express
physical layer receiver includes: a serializer/deserializer
(SERDES) configured to convert serial data received through each of
the lanes to parallel data of about 10 bits; a Physical Layer
Device (PHY) Interface for the PCI Express Architecture (PIPE)
configured to perform a bit alignment process on the 10-bit
parallel data using the comma symbol, configured to convert the
10-bit parallel data to 8-bit parallel data, and configured to add
the skip symbol to the received data or remove the skip symbol from
the received data based on a status of the received data; and a
skew compensator configured to set the alignment point of the 8-bit
received data corresponding to each of the lanes based on the comma
symbol included in a training sequence, and configured to
compensate for the byte skew between the received data by shifting
the alignment point of the received data corresponding to each of
the lanes by reflecting an addition or a removal of the skip
symbol. The serializer/deserializer may be a deserializer, and may
convert serial data to parallel data. The training sequence may
include TS1 and TS2 patterns of the PCI Express bus. The training
sequence may be a test pattern of about 2.5 Gb/s. Several training
sequences TS1 and TS2 may be received during an initialization
process of the PCI Express bus.
In further embodiments, the comma symbol may be an 8-bit symbol
having a predetermined bit composition used for compensating for
the byte skew between the lanes. In an example embodiment of the
present invention, the addition or the removal of the skip symbol
may be performed in the PIPE of the PCI Express physical layer
receiver, and the byte skew between the lanes may be effectively
compensated for by reflecting the addition or the removal of the
skip symbol on the shift operation of the alignment point. The
number of the lanes of the PCI Express bus may be 4.
The skew compensator may compensate for the byte skew between the
received data through the lanes based on the method of compensating
for the byte skew of the PCI Express bus.
Consequently, the byte skew of the PCI Express bus may be
effectively compensated for despite the addition or the removal of
the skip symbol.
BRIEF DESCRIPTION OF THE DRAWINGS
The above and other features and advantages of the present
invention will become more apparent when described in detailed
example embodiments thereof with reference to the attached drawings
in which:
FIG. 1 is a block diagram illustrating a data transmission of a
conventional PCI Express bus;
FIG. 2 is a block diagram illustrating a data reception of a
conventional PCI Express bus;
FIG. 3 is a timing diagram illustrating a conventional method of
compensating for a byte skew using only comma symbol;
FIG. 4 is a flowchart illustrating a method of compensating for a
byte skew of PCI Express bus according to an example embodiment of
the present invention;
FIG. 5 is a flowchart illustrating an example embodiment of a step
S420 shown in FIG. 4;
FIG. 6 is a flowchart illustrating an example embodiment of a step
S520 shown in FIG. 5;
FIG. 7 is a flowchart illustrating an example embodiment of a step
S430 shown in FIG. 4;
FIGS. 8A and 8B are timing diagrams illustrating an alignment point
setting process according to an example embodiment of the present
invention;
FIG. 9 is a timing diagram illustrating an alignment point shift
process according to an example embodiment of the present
invention; and
FIG. 10 is a block diagram illustrating a PCI Express physical
layer receiver according to an example embodiment of the present
invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Detailed illustrative embodiments of the present invention are
disclosed herein. However, specific structural and functional
details disclosed herein are merely representative for purposes of
describing example embodiments of the present invention. This
invention may, however, be embodied in many alternate forms and
should not be construed as limited to the embodiments set forth
herein.
Accordingly, while the invention is susceptible to various
modifications and alternative forms, specific embodiments thereof
are shown by way of example in the drawings and will herein be
described in detail. It should be understood, however, that there
is no intent to limit the invention to the particular forms
disclosed, but on the contrary, the invention is to cover all
modifications, equivalents, and alternatives falling within the
spirit and scope of the invention. Like numbers refer to like
elements throughout the description of the figures.
It will be understood that, although the terms first, second, etc.
may be used herein to describe various elements, these elements
should not be limited by these terms. These terms are only used to
distinguish one element from another. For example, a first element
could be termed a second element, and, similarly, a second element
could be termed a first element, without departing from the scope
of the present invention. As used herein, the term "and/or"
includes any and all combinations of one or more of the associated
listed items.
It will be understood that when an element is referred to as being
"connected" or "coupled" to another element, it can be directly
connected or coupled to the other element or intervening elements
may be present. In contrast, when an element is referred to as
being "directly connected" or "directly coupled" to another
element, there are no intervening elements present. Other words
used to describe the relationship between elements should be
interpreted in a like fashion (i.e., "between" versus "directly
between", "adjacent" versus "directly adjacent", etc.).
The terminology used herein is for the purpose of describing
particular embodiments only and is not intended to be limiting of
the invention. As used herein, the singular forms "a", "an" and
"the" are intended to include the plural forms as well, unless the
context clearly indicates otherwise. It will be further understood
that the terms "comprises", "comprising", "includes" and/or
"including", when used herein, specify the presence of stated
features, integers, steps, operations, elements, and/or components,
but do not preclude the presence or addition of one or more other
features, integers, steps, operations, elements, components, and/or
groups thereof.
Unless otherwise defined, all terms (including technical and
scientific terms) used herein have the same meaning as commonly
understood by one of ordinary skill in the art to which this
invention belongs. It will be further understood that terms, such
as those defined in commonly used dictionaries, should be
interpreted as having a meaning that is consistent with their
meaning in the context of the relevant art and will not be
interpreted in an idealized or overly formal sense unless expressly
so defined herein.
It should also be noted that in some alternative implementations,
the functions/acts noted in the blocks may occur out of the order
noted in the flowcharts. For example, two blocks shown in
succession may in fact be executed substantially concurrently or
the blocks may sometimes be executed in the reverse order,
depending upon the functionality/acts involved.
Particularly, data may be transmitted in a form of a stream, so
"data" may sometimes be understood as "a data stream" or "data
streams".
FIG. 4 is a flowchart illustrating a method of compensating for a
byte skew of PCI Express bus according to an example embodiment of
the present invention.
Referring to FIG. 4, the method includes a step S410 of determining
whether or not received data are a training sequence.
The training sequence includes TS1 and TS2 patterns of the PCI
Express bus. For example, the training sequence may be a test
pattern having a clock frequency of about 2.5 Gb/s. Several
training sequences TS1 or TS2 pattern may be received during an
initialization process of the PCI Express bus.
When the received data are the training sequence, an alignment
point corresponding to each of the lanes is set based on a comma
symbol included in the training sequence at step S420.
The comma symbol may be an 8-bit symbol having a predetermined bit
composition. The number of the lanes may be four.
The received data are stored in a register corresponding to each of
the lanes. For example, the register may be a FIFO (First In, First
Out) type shift register that is capable of storing the 8-bit data
for 5 clocks.
A depth of the register may be defined as 5 when the shift register
stores data for 5 clocks. When the byte skew between the lanes is a
maximum of 4 clocks, the byte skew between the lanes may be
sufficiently compensated for since the depth of the register is
5.
The alignment point may be a delay point of the shift register
corresponding to each of the lanes. That is, when the shift
register, whose depth is 5, is used for each of the lanes, the
alignment point may correspond to the delay points 1 through 5.
The alignment point corresponding to the delay point 1 may be set
corresponding to the latest stored data, and the alignment point
corresponding to the delay point 5 is set corresponding to
four-clocks delayed data on the latest stored data.
FIGS. 8A and 8B are timing diagrams illustrating an alignment point
setting process according to an example embodiment of the present
invention.
In FIGS. 8A and 8B, data symbols represented as FE, COM, 0, 1, 2,
3, 4, 5, 6, 7, 8, 9, 10, 11 and 12 may be 8-bit data symbols,
respectively.
FIG. 8A is a timing diagram illustrating data received through four
lanes before the alignment point setting process is performed.
Referring to FIG. 8A, a byte skew exists between data received
through four lanes L0_RX_DATA, L1_RX_DATA, L2_RX_DATA and
L3_RX_DATA before the alignment point setting process based on the
comma symbol is performed.
Accordingly, the alignment point is to be set on each of the lanes,
and the data received through each of the lanes are to be aligned
based on the set alignment point.
FIG. 8B is a timing diagram illustrating received data aligned
based on the set alignment point according to an example embodiment
of the present invention.
Referring to FIG. 8B, it is shown that the data received through
each of the lanes L0_RX_DATA, L1_RX_DATA, L2_RX_DATA and L3_RX_DATA
are aligned based on the set comma symbol, respectively.
When the 8-bit register, whose depth is 5, is used for each of the
lanes, a status of the register for each of the lanes corresponding
to the timing diagram shown in FIG. 8A may be represented as Table
1 through Table 4.
TABLE-US-00001 TABLE 1 L0_RX_DATA REGISTER ST0 ST1 ST2 ST3 ST4 t =
N FE FE FE FE FE t = N + 1 COM FE FE FE FE t = N + 2 0 COM FE FE FE
t = N + 3 4 0 COM FE FE t = N + 4 8 4 0 COM FE
TABLE-US-00002 TABLE 2 L1_RX_DATA REGISTER ST0 ST1 ST2 ST3 ST4 t =
N FE FE FE FE FE t = N + 1 FE FE FE FE FE t = N + 2 FE FE FE FE FE
t = N + 3 COM FE FE FE FE t = N + 4 1 COM FE FE FE
TABLE-US-00003 TABLE 3 L2_RX_DATA REGISTER ST0 ST1 ST2 ST3 ST4 t =
N COM FE FE FE FE t = N + 1 2 COM FE FE FE t = N + 2 6 2 COM FE FE
t = N + 3 10 6 2 COM FE t = N + 4 12 10 6 2 COM
TABLE-US-00004 TABLE 4 L3_RX_DATA REGISTER ST0 ST1 ST2 ST3 ST4 t =
N FE FE FE FE FE t = N + 1 FE FE FE FE FE t = N + 2 COM FE FE FE FE
t = N + 3 3 COM FE FE FE t = N + 4 7 3 COM FE FE
Referring to Table 1 through Table 4, at time `t=N`, the comma
symbol COM is first detected in the data received through the lane
L2_RX_DATA as shown in FIG. 8A.
Other times `t=N+1`, `t=N+2`, `t=N+3` and `t=N+4` denote one clock
delayed timing of time `t=N`, two-clocks delayed timing of time
`t=N`, three-clocks delayed timing of time `t=N` and four-clocks
delayed timing of time `t=N`, respectively.
As shown in Table 1 through Table 4, `ST0`, `ST1`, `ST2`, `ST3` and
`ST4` denote a first stage, a second stage, a third stage, a fourth
stage and a fifth stage of the shift register.
As shown in Table 1 through Table 4 and FIGS. 8A through 8B, `COM`
denotes the comma symbol, Arabic numerals denote a data symbol and
`FE` denotes a symbol transferred in advance of the comma symbol,
and thus, is not a matter of concern.
Table 1 represents a status of the register L0_RX_DATA REGISTER
corresponding to the lane L0_RX_DATA. The register L0_RX_DATA
REGISTER corresponding to the lane L0_RX_DATA is a FIFO type shift
register, and shifts input data at each clock to store the shifted
input data.
The input data stored in the shift register at each clock are
shifted from the stage ST0 to the stage ST4. Since the depth of the
shift register is 5, the shift register may store the shifted input
data for 5 clocks.
As shown in Table 1, the comma symbol COM of the data received
through the lane L0_RX_DATA is stored in the shift register at time
`t=N+1`, one clock later than time `t=N`, when the comma symbol is
first stored in the shift register corresponding to the lane
L2_RX_DATA.
The comma symbol COM of the data received through the lane
L0_RX_DATA is shifted by 1 stage until time `t=N+4`, four clocks
later than time `t=N`, when the comma symbol is first stored in the
shift register corresponding to the lane L2_RX_DATA.
Table 2 represents the status of the register L1_RX_DATA REGISTER
corresponding to the lane L1_RX_DATA. The register L1_RX_DATA
REGISTER corresponding to the lane L1_RX_DATA is a FIFO type shift
register, and shifts input data at each clock to store the shifted
input data.
The input data stored in the shift register at each clock are
shifted from the stage ST0 to the stage ST4. Since the depth of the
shift register is 5, the shift register may store the shifted input
data for 5 clocks.
As shown in Table 2, the comma symbol COM of the data received
through the lane L1_RX_DATA is stored in the shift register at time
`t=N+3`, three clocks later than time `t=N`, when the comma symbol
is first stored in the shift register corresponding to the lane
L2_RX_DATA.
The comma symbol COM of the data received through the lane
L1_RX_DATA is shifted by 1 stage until time `t=N+4`, four clocks
later than time `t=N`, when the comma symbol is first stored in the
shift register corresponding to the lane L2_RX_DATA.
Table 3 represents the status of the register L2_RX_DATA REGISTER
corresponding to the lane L2_RX_DATA. The register L2_RX_DATA
REGISTER corresponding to the lane L2_RX_DATA is a FIFO type shift
register, and shifts input data at each clock to store the shifted
input data.
The input data stored in the shift register at each clock are
shifted from the stage ST0 to the stage ST4. Because the depth of
the shift register is 5, the shift register may store the shifted
input data for 5 clocks.
As shown in Table 3, the comma symbol COM of the data received
through the lane L2_RX_DATA is stored in the shift register at time
`t=N`. The comma symbol COM of the data received through the lane
L2_RX_DATA is shifted by 1 stage until time `t=N+4`, four clocks
later than time `t=N`, when the comma symbol is first stored in the
shift register corresponding to the lane L2_RX_DATA.
Table 4 represents the status of the register L3_RX_DATA REGISTER
corresponding to the lane L3_RX_DATA. The register L3_RX_DATA
REGISTER corresponding to the lane L3_RX_DATA is a FIFO type shift
register, and shifts input data at each clock to store the shifted
input data.
The input data stored in the shift register at each clock are
shifted from the stage ST0 to the stage ST4. Because the depth of
the shift register is 5, the shift register may store the shifted
input data for 5 clocks.
As shown in Table 4, the comma symbol COM of the data received
through the lane L3_RX_DATA is stored in the shift register at time
`t=N+2`, two clocks later than time `t=N`, when the comma symbol is
first stored in the shift register corresponding to the lane
L2_RX_DATA.
The comma symbol COM of the data received through the lane
L3_RX_DATA is shifted by 1 stage until time `t=N+4`, four clocks
later than time `t=N`, when the comma symbol is first stored in the
shift register corresponding to the lane L2_RX_DATA.
At time `t=N+3`, when all of the comma symbols corresponding to
each of the four lanes are stored in their corresponding shift
register, the alignment point is set based on the stage of the
shift register where the comma symbol is stored.
That is, in the shift register L0_RX_DATA REGISTER corresponding to
the lane L0_RX_DATA shown in Table 1, and at time `t=N+3`, when all
of the comma symbols corresponding to each of the four lanes are
stored in their corresponding shift register, the comma symbol may
be shifted to the third stage ST2 from the first stage ST0 by two
stages. Accordingly, for example, the alignment point of the lane
L0_RX_DATA may be set as a binary value `00111`.
In the shift register L1_RX_DATA REGISTER corresponding to the lane
L1_RX_DATA shown in Table 2, and at time `t=N+3`, when all of the
comma symbols corresponding to each of the four lanes are stored in
their corresponding shift register, the comma symbol may be stored
in the first stage ST0. Accordingly, for example, the alignment
point of the lane L1_RX_DATA may be set as a binary value
`00001`.
In the shift register L2_RX_DATA REGISTER corresponding to the lane
L2_RX_DATA shown in Table 3, and at time `t=N+3`, when all of the
comma symbols corresponding to each of the four lanes are stored in
their corresponding shift register, the comma symbol may be shifted
to the fourth stage ST3 from the first stage ST0 by three stages.
Accordingly, for example, the alignment point of the lane
L2_RX_DATA may be set as a binary value `01111`.
In the shift register L3_RX_DATA REGISTER corresponding to the lane
L3_RX_DATA shown in Table 4, and at time `t=N+3`, when all of the
comma symbols corresponding to each of the four lanes are stored in
their corresponding shift register, the comma symbol may be shifted
to the second stage ST1 from the first stage ST0 by one stage.
Accordingly, for example, the alignment point of the lane
L3_RX_DATA may be set as a binary value `00011`.
As a result, a transmission time of data, after the comma symbol,
that are received by each of the four lanes may be synchronized by
setting the alignment point of each of the four lanes based on the
comma symbol.
In the timing diagram shown in FIG. 8A, when the data `0` received
by the lane L0_RX_DATA are delayed by two clocks, the data `1`
received by the lane L1_RX_DATA are not delayed, the data `2`
received by the lane L2_RX_DATA are delayed by three clocks and the
data `3` received by the lane L3_RX_DATA are delayed by one clock,
all of the data received through each of the four lanes may be
aligned based on the comma symbol as shown in the timing diagram of
FIG. 8B.
Referring back to FIG. 4, when the received data are not in the
training sequence, the alignment point is shifted by reflecting an
addition or a removal of the skip symbol on the data received
through each of the four lanes at step S430.
When the PCI Express transmitter transmits the skip symbols with
data symbols, the PCI Express receiver may add the skip symbols to
the received data or may remove the skip symbols from the received
data so as to prevent an overflow or underflow of an elastic buffer
included in the PCI Express receiver.
Since the skip symbol is received after the comma symbol is
received, in order to correctly compensate for the byte skew
between the data received through each of the four lanes, the
alignment process reflecting the addition or the removal of the
skip symbol on a shift operation of the alignment point should be
selectively performed as well as the alignment process using the
comma symbol.
FIG. 9 is a timing diagram illustrating an alignment point shift
process according to an example embodiment of the present
invention.
Referring to FIG. 9, after the alignment process based on the comma
symbol is performed, the skip symbol is removed from the data
received through the lane L1_RX_DATA, and the skip symbol is added
to the data received through the lane L2_RX_DATA.
Despite the alignment process based on the comma symbol, the byte
skew occurs between the data received through each of the four
lanes L0_RX_DATA, L1_RX_DATA, L2_RX_DATA and L3_RX_DATA due to the
addition or the removal of the skip symbol. Accordingly, in order
to align the data received through each of the four lanes, each of
the alignment points is to be shifted, respectively.
That is, when the skip symbol is added to the received data, the
corresponding alignment point is adjusted so that the received data
are less delayed, and when the skip symbol is removed from the
received data, the corresponding alignment point is adjusted so
that the received data are more delayed.
Based on the comma symbol as shown in FIG. 9, when the alignment
point of the lane L0_RX_DATA is set as the binary value `00111`,
the alignment point of the lane L1_RX_DATA is set as the binary
value `00001`, the alignment point of the lane L2_RX_DATA is set as
the binary value `01111` and the alignment point of the lane
L3_RX_DATA is set as the binary value `00011`, the alignment point
of the lane L1_RX_DATA is shifted from the binary value `00001` to
`00011`, and the alignment point of the lane L2_RX_DATA is shifted
from the binary value `01111` to `00111` since the skip symbol is
removed from the data received through the lane L1_RX_DATA, and the
skip symbol is added to the data received through the lane
L2_RX_DATA.
The reason why the alignment point of the lane L1_RX_DATA is
shifted from the binary value `00001` to `00011` is for more
delaying the data received through the lane L1_RX_DATA.
The reason why the alignment point of the lane L2_RX_DATA is
shifted from the binary value `01111`to `00111` is for less
delaying the data received through the lane L2_RX_DATA.
Each of the alignment points varied due to the addition or removal
of the skip symbol may be adjusted as described above.
Consequently, the byte skew between the data received through each
of the four lanes may be effectively compensated for by selectively
aligning the data received through each of the four lanes.
FIG. 5 is a flowchart illustrating an example embodiment of step
S420 shown in FIG. 4.
Referring to FIG. 5, the step S420 shown in FIG. 4 includes step
510 of determining whether or not the lane where the comma symbol
is detected exists.
A training sequence, such as the TS1 and TS2 patterns of the PCI
Express bus, carries the comma symbol at a predetermined interval.
Accordingly, in step S510, it is determined whether or not the lane
where the comma symbol included in the training sequence is
detected exists.
When the lane where the comma symbol included in the training
sequence is detected exists, it is determined at step S520 whether
or not the comma symbols are detected in all of the lanes during a
predetermined time period.
The predetermined time period is associated with the depth of the
register corresponding to each of the lanes. That is, when the
depth of the register is 5, it is determined whether the comma
symbols are detected in all of the lanes within 5 clocks. The
reason for determining whether or not the comma symbols are
detected in all of the lanes within the 5 clocks when the depth of
the register is 5, is that when the 5 clocks have elapsed, the
register, where the comma symbol is first detected, no longer
stores the comma symbol.
When the comma symbols are detected in all of the lanes during the
predetermined time period, each of the alignment points is set at
step S530 based on each of the timings when each of the comma
symbols is detected.
The alignment point may be a delay point of the shift register
corresponding to each of the lanes. That is, when the shift
register having the depth of 5 is used for each of the lanes, the
alignment point may correspond to delay points 1 through 5.
The alignment point corresponding to the delay point 1 corresponds
to the latest stored data, and the alignment point corresponding to
the delay point 5 corresponds to the four-clocks delayed data on
the latest stored data.
When the comma symbols are not detected in all of the lanes during
the predetermined time period, comma symbol detection information
is reset at step S540.
The comma symbol detection information may contain information on
the lanes where the comma symbol has been detected previously.
When the comma symbols are not detected in all of the lanes during
the predetermined time period, a comma symbol detection process on
all of the lanes is re-started.
Accordingly, the comma symbol detection information containing the
information on the lanes where the comma symbol has been detected
previously is reset and the comma symbol detection process may be
newly performed.
The comma symbol detection information includes all of the
information required for determining whether or not the comma
symbols are detected in all of the lanes. For example, the comma
symbol detection information may include information on which stage
of the register that corresponds to the lane is where the comma
symbol is stored. After the comma symbol detection information is
reset, the process flow of step S420 returns to step S510.
FIG. 6 is a flowchart illustrating an example embodiment of step
S520 shown in FIG. 5.
Referring to FIG. 6, step S520 includes step S610 of starting a
timer when the lane where the comma symbol is received exists. The
timer is for measuring a time period associated with the depth of
the register corresponding to each of the lanes.
When the depth of the register is 5, it is determined whether or
not the comma symbols are detected in all of the lanes within the 5
clocks. The reason for determining within the 5 clocks whether or
not the comma symbols are detected in all of the lanes when the
depth of the register is 5, is that when the 5 clocks have elapsed,
the register, where the first comma symbol is detected, no longer
stores the comma symbol. Accordingly, the timer may measure a time
period corresponding to the 5 clocks.
Then, it is determined at step S620 whether or not the comma
symbols are detected in all of the lanes. Step 520 also includes
step S630 of determining whether or not the timer is expired.
Step S620 and step S630 may be performed in the order shown in FIG.
6, or may be performed in reverse order from that shown in FIG. 6,
or may be performed at the same time.
Before the timer is expired, when the comma symbol is received by
all of the lanes, the process flow returns to step S530 shown in
FIG. 5, and then, each of the alignment points is set based on each
of the timings when each of the comma symbols is detected.
When the comma symbol is not received by all of the lanes and the
timer is not expired, the process flow returns to step S620 and
then, it is determined whether or not the comma symbol is received
by all of the lanes.
When the comma symbol is not received by all of the lanes and the
timer is expired, the process flow returns to step S540 shown in
FIG. 5 and then, the comma symbol detection information is
reset.
FIG. 7 is a flowchart illustrating an example embodiment of step
S430 shown in FIG. 4.
Referring to FIG. 7, step S430 shown in FIG. 4 includes step S710
of determining whether or not the skip symbol is added to the
received data or is removed from the received data.
It may be determined whether the skip symbol is added to or is
removed from the received data using information on the addition or
the removal of the skip symbol provided from a PHY Interface for
the PCI Express Architecture (PIPE) of the PCI Express bus.
When the skip symbol is added to the received data, it is
determined at step S720 whether or not the alignment point of the
corresponding received data may be shifted so that the
corresponding received data are less delayed.
The reason for determining whether or not the alignment point of
the corresponding received data may be shifted is that there may
exist a case in which the alignment point of the corresponding
received data may not be shifted, so that the corresponding
received data are less delayed when the skip symbol is added to the
received data as described in Table 2.
When the received data may be less delayed, the alignment point of
the corresponding received data is shifted at step S740 so that the
received data are less delayed.
For example, as described in FIG. 9, when the skip symbol is added
to the received data, the alignment point is shifted from the
binary value `01111` to `00111`.
When the received data cannot be less delayed, each of the
alignment points of other lanes is shifted at step S760 so that the
received data corresponding to the other lanes are more
delayed.
That is, in case the lane where the received data cannot be less
delayed exists, each of the alignment points may be appropriately
adjusted by considering whether or not other lanes, except for the
lane where the received data cannot be less delayed, receive the
data to which the skip symbol is added or from which the skip
symbol is removed.
The depth of the shift register corresponding to each of the lanes
should be suitably set, so that a condition, in which a shift
operation of the alignment point is impossible, cannot occur.
When the skip symbol is removed from the received data, it is
determined at step S730 whether or not the alignment point of the
corresponding received data may be shifted so that the
corresponding received data are more delayed.
The reason for determining whether or not the alignment point of
the corresponding received data may be shifted is that there may
exist a case in which the alignment point of the corresponding
received data cannot be shifted, so that the corresponding received
data are more delayed when the alignment point has the binary value
such as `11111` even though the skip symbol has been removed from
the received data.
When the received data may be more delayed, the alignment point of
the corresponding received data is shifted at step S750 so that the
received data are more delayed.
For example, as described in FIG. 9, when the skip symbol is
removed from the received data, the alignment point is shifted from
the binary value `00001` to `00011`.
When the received data cannot be more delayed, each of the
alignment points of other lanes is shifted at step S770 so that the
received data corresponding to the other lanes are less
delayed.
That is, in case the lane where the received data cannot be more
delayed exists, each of the alignment points may be appropriately
adjusted by considering whether or not other lanes, except for the
lane where the received data may not be more delayed, receive the
data to which the skip symbol is added or from which the skip
symbol is removed.
The depth of the shift register corresponding to each of the lanes
should be suitably set, so that a condition, in which a shift
operation of the alignment point is impossible, cannot occur.
FIG. 10 is a block diagram illustrating a PCI Express physical
layer receiver according to an example embodiment of the present
invention.
Referring to FIG. 10, the PCI Express physical layer receiver
includes a serializer/deserializer 910, a PHY Interface for the PCI
Express Architecture (PIPE) 920, and a skew compensator 930.
The SERDES 910 converts serial data received through each of the
lanes to 10 bits of parallel data. The serializer/deserializer 910
may be a deserializer.
The PIPE 920 performs a bit alignment process on the 10-bit
parallel data using the comma symbol, converts the 10-bit parallel
data to 8-bit parallel data, and adds the skip symbol to the
received data or removes the skip symbol from the received data
based on a status of the received data.
The PIPE 920 may include an elastic buffer, may remove the skip
symbol from the received data so as to prevent an overflow of the
elastic buffer, and may add the skip symbol to the received data so
as to prevent an underflow of the elastic buffer.
The PIPE 920 may be defined by the PIPE specification of Intel
Corporation. The PIPE specification of Intel Corporation is
designed for developing a functionally equivalent PCI Express
physical layer.
The skew compensator 930 sets the alignment point of the received
data corresponding to each of the lanes based on the comma symbol
included in the training sequence, and compensates for the byte
skew between the received data by shifting the alignment point of
the received data corresponding to each of the lanes according to
an addition or a removal of the skip symbol. The skew compensator
930 may receive information on the addition or the removal of the
skip symbol from the PIPE 920.
The training sequence includes TS1 and TS2 patterns of the PCI
Express bus. For example, the training sequence may be a test
pattern having a clock frequency of about 2.5 Gb/s. Several
training sequences such as the TS1 and/or TS2 patterns may be
received during an initialization process of the PCI Express bus.
The comma symbol may be an 8-bit symbol having a predetermined bit
composition, which is used for compensating for the byte skew
between the lanes. For example, the number of the lanes may be
4.
The skew compensator 930 compensates for the byte skew between the
received data through each of the lanes based on the method of
compensating for the byte skew of the PCI Express bus as described
in FIGS. 4 through 7.
The skew compensator 930 may be implemented by programming a micro
controller so that the method described in FIGS. 4 through 7 is
performed.
The skew compensator 930 may be implemented by coding in VHDL (Very
High Speed Integrated Circuit Hardware Description Language) and
performing a synthesis process.
As described above, the method of compensating for the byte skew
between the received data corresponding to each of the lanes and
the PCI Express physical layer receiver set the alignment point of
the received data corresponding to each of the lanes using the
comma symbol included in the training sequence, and reflect the
addition or removal of the skip symbol on the shift operation of
the alignment point.
Therefore, the method and the PCI Express physical layer receiver
may effectively compensate for the byte skew between the received
data corresponding to each of the lanes.
Further, the PCI Express receiver may quickly and correctly recover
the received data.
While the example embodiments of the present invention and their
advantages have been described in detail, it should be understood
that various changes, substitutions and alterations may be made
herein without departing from the scope of the invention.
* * * * *